MetaInsight

Network meta-analysis can help determine the “best” treatment option

Complex reviews require extensive statistical and programming knowledge

• Bewildering R packages:
  • BUGSnet, bnma, coda, gemtc, meta, metafor, netmeta
• Some dependent on JAGS which may be hard to install
• Require data in different formats and use different terminology

Shiny removes barriers for accessing cutting-edge methods

Lack of reproducibility can limit uptake

• {shinyscholar} was forked from {wallace} to make development of reproducible apps easier
• Convert core functionality into functions and package
• The app becomes an interface to the functions, purely focused on interactivity

Reproducibility relies on a strict structure

• Each module has an id made up of the component and module e.g. summary_network
• The module calls a synonymous function summary_network()
• Input values are stored in common$meta$summary_network$<input id>
• Values are knitted into an .Rmd chunk and combined to create a .qmd

Reproducibility relies on a strict structure

```{asis, echo = {{summary_network_knit}, eval = {{summary_network_knit}}, include = {{summary_network_knit}}}}
### Display the networks for the original data and data with excluded studies.
```
```{asis, echo = {{summary_network_knit}, eval = {{summary_network_knit}}, include = {{summary_network_knit}}}}
{r,  results = 'asis'}
```
```{r, echo = {{summary_network_knit}, include = {{summary_network_knit}}}}
network_style <- {{summary_network_style}}

summary_network_all <- summary_network(frequentist_all, bugsnet_all, network_style, {{summary_network_label_all}}, "Network plot of all studies")
htmltools::HTML(summary_network_all)
```

Reproducibility relies on a strict structure

### Display the networks for the original data and data with excluded studies.

```{r, results = 'asis'}
network_style <- "netplot"

summary_network_all <- summary_network(frequentist_all, bugsnet_all, network_style, 1, "Network plot of all studies")
htmltools::HTML(summary_network_all$svg)
```

Reproducibility also enables improved reporting

Risk of bias

• MetaInsight enables sensitivity analyses by excluding studies
• During reviews, risk of bias information can be collected e.g.
  • Randomisation, blinding, missing data
• Users can now upload these scores and use them to guide sensitivity analyses

Integration with CiNEMA helps to evaluate

Acknowledgments

• Tom Morris, Janion Nevill, Ryan Field, Alex Sutton, Nicola Cooper
• Wellcome
• NIHR
• Chan Zuckerburg Initiative